MoCam

MoCam: Video Re-Camera via Structured Denoising Dynamics

Orange Team
Haofeng Liu, Yang Zhou, Ziheng Wang, Zhengbo Xu, Zhan Peng, Jie Ma, Jun Liang, Shengfeng He, Jing Li
Paper Released! GitHub (Coming Soon) 🤗 HF Model (Coming Soon)

TL;DR: We propose MoCam, a method that enables robust novel view synthesis even with highly incomplete or distorted geometric priors. MoCam introduces structured denoising dynamics that guide the diffusion process from initial geometric alignment to final appearance refinement, producing coherent and photorealistic results for both static and dynamic scenes.

place gif

Demo

Large-movement Trajectories

Complex Trajectories

Dolly Zoom

Zoom In/Out

Bullet Time

BibTeX

@misc{orange2025mocam,
  title={MoCam: Unified Novel View Synthesis via Structured Denoising Dynamics}, 
  author={Haofeng Liu, Yang Zhou, Ziheng Wang, Zhengbo Xu, Zhan Peng, Jie Ma, Jun Liang, Shengfeng He, Jing Li},
  journal={arXiv preprint arXiv:2605.12119},
  year={2026},
  url={https://orange-3dv-team.github.io/MoCam}, 
}